How do I crop a medical images based on selective search automatically.

i need to crop a particulr region of medical ultrasound images auomaticaly.

[Walter Roberson]

Is it a fixed position inside the image?

If not then does it have distinguishing characteristics?

[THUNAKALA BALA KRISHNA]

thanks for response.

No fixed position inside the image. i am having ultrasound images of diffrent organs. i need to crop those organs auomaticaly.

[KSSV]

Attach one image and show us your region of interest.

[Walter Roberson]

To do that, you will need to use machine learning. You will need at least 5 different labeled images for each different state of each different organ that you need to deal with. "Labeled" in this sense meaning that you will need to have images and you will need to have had an expert trace around the organ, with the trace stored separately. Each different "state" in the sense that if, for example, someone had an enlarged spleen, you would need examples of different ways the enlargement can occur; if someone's liver wasn't attached completely properly, you would need pictures of that and those pictures would have to give examples of the boundary to decide what was to be considered part of the liver itself and what was to be excluded.

Computers are not good on generalizing without examples. If you were to train only on normal spleen, and then to try to have it work on enlarged spleens, then it would either not be recognized or too much would be cropped or too little would be cropped, or it might be recognized as a different organ...

Remember to train on examples of the organs with gunshot wounds...

Answer 1

[yanqi liu]

sir，may be upload some image file to analysis，aslo can use roipoly to segment and crop

[Walter Roberson]

roipoly() does not work until you have identified the area to create the Region of Interest (ROI) around. The user needs a way of automatically detecting the boundaries of any of a number of different organs in ultrasound images -- including cases where the organs are not in their normal shape or the organs are diseased.